Ramen fans can now easily identify different bowls of ramen served by Ramen Jiro. This is by using a custom of Google’s AutoML Vision. Kenji Doi, a data scientist, collected over 48,000 photos of ramen from Ramen Jiro’s 41 shops. Ramen Jiro is a chain of ramen shops that is based in Tokyo.

Doi uploaded these photos to AutoML, which took one whole day to analyze and classify all the data. It was able to accurately identify the menu item about 95% of the time. It bases its predictions on the cuts of the meat used and the placement of the toppings on the bowl.